- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Fri, 20 Apr 2007 10:58:20 +0100
- To: Bijan Parsia <bparsia@cs.man.ac.uk>
- CC: Michael Schneider <m_schnei@gmx.de>, public-owl-dev@w3.org
Jena Models have a reification style, and at least some variants keep partially reified triples in a special place, so that when all the other components turn up it can all be put together. Chris Dollin handles that bit of the implementation (lucky for me!) Overall, we've spent quite a lot of time and effort supporting reification, despite its problems; and at least some users expect us to continue to do so. At various stages we toy with the idea of simply treating reification as an I/O issue (i.e. the parser and serializer know the rdf:ID construct, and you get five triples not one); but I keep getting out-voted and we continue to have some magic that allows one triple to be stored with a reification ID. I think it is probably correct that I get outvoted. I think too much of the lack of elegance of the solution; the others are listening better to the user-base. Jeremy Bijan Parsia wrote: > > On 19 Apr 2007, at 11:03, Bijan Parsia wrote: > >> >> On 19 Apr 2007, at 10:50, Jeremy Carroll wrote: >> >>> Bijan Parsia wrote: >>>> (I'm not aware, for example, of any toolkit which maps reified >>>> triples into compact form the way e.g, CWM does with the list >>>> vocabulary). >>> >>> Jena does. >> >> That's interesting! I didn't know that. >> >> How do you handle incomplete reifictations, e.g., missing an >> rdf:subject or the like? >> >> Is this in all models, or just in some? > > By which, of course, I mean "Jena Models". > > Cheers, > Bijan. > -- Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Friday, 20 April 2007 09:59:30 UTC